(Updates made 11/8) With news this week from “phosphor solution provider” Intematix
that they had achieved a 203 lumen/Watt light engine using a remote phosphor approach
(story here),
it would seem to be a decent opportunity to shine the proverbial spotlight in
the direction of those remote and “cold” phosphor solutions. It always
strikes me as interesting that there aren’t more remote phosphor implementations,
in the sense that so much of LED lighting has, so far, been about replicating
how lighting has been done in the past. One could argue the semantics of it a
bit, but the legacy approach that stems from fluorescent lamps is to coat a big
surface (glass tube) with a combination of phosphors, then excite it with lower
wavelength light, specifically an electric arc generating UV light. Since a fluorescent
tube, whether straight or twisty, shoots that arc from one end to the other, it’s
tricky to say it’s purely remote, but if we projected the concept and imagined
LEDs shooting blue or violet light “into the tube”, we’d clearly call
that a remote approach, since the phosphor is clearly separate from the emitters.

There
are pluses and minuses to every approach, and it’s no different here. On the plus
side, physically separating the phosphor coating from the LED emitter removes
it from the substantial heat that is present on the actual LED. Trying not to
get too geeky here, but I’m a believer that little doses of “how that works”
help us to make sense of bigger pictures, such as “why thermal management
is so important in LED systems”. Doing a quick thumbnail on how hot that
LED is, if we assume a 1 watt LED in a 3535 package (3.5mm x 3.5mm), at 50% conversion
efficiency, we come up with something like 25 watts/sq-in being generated in the
form of heat. It doesn’t go into the beam, but rather has to spread over the emitter
and conduct out the back side. In terms of a comparison most of us would get,
if we have an electric grill in the kitchen that’s about a 12 x 18 inches (216
square inches), at 25 watts/sq-in, that would be like juicing it with about 5000
watts. The grill in our kitchen uses only about 1000 watts, and takes about 10
seconds from off to ouch if you hold your hand on it.

While phosphor itself
is fairly bulletproof, the stuff that binds it all together, especially any organic
materials, aren’t immune to heat effects. Since LEDs don’t have heat in the beam,
if you shine blue LEDs at a phosphor that’s not physically attached to those emitters,
the only heat it has to deal with results from the conversion process itself which
is a result of photons moving from higher energy states (shorter wavelengths,
like blue) to lower ones (green, yellow, orange, red, in that reverse ROY G BIV
order). The further the spectral jump, the more energy that photon has to give
up, and where does the energy go? It becomes heat, with a greater effect in the
bigger jump. One of the results of that is the notorious “red fade”
in which the silicone binder in that phosphor blend tends to turn a bit brown
as a result of the heat, with more of that browning occuring next to the red phosphor
elements, where more heat is being released from that “Stokes shift”.
That sort of causes a bit of a cataracts kind of deal, resulting in less red being
generated as the LED ages. There are some substitute materials, such as glass,
that can be used in the phosphor blend, which naturally begins a cost-benefit
kind of trade-off. While the conversion-related heat isn’t insignificant, both
the remote and direct contact phosphors experience the same effect. The advantage
to the remote approach is that the conversion heat is spread out over a much wider
area, which would be of help with the longevity of those added materials, helping
to minimize the color shift (Xicato, for example, has made this a big point in
promoting the benefits of their “corrected cold phosphor” LED modules.
In the direct application white LED approach, the materials are being subjected
to the additional heat from the LED itself, plus the LED has to deal with some
additional heat added in there as a result of what the phosphor is generating.
Both conspire to make it more difficult to maintain a specific color point over
time. (A big thank you to Steve Paolini, CTO with NEXT
Lighting for taking the initiative to help educate me on the intricacies of
this phosphor conversion process, resulting in some updates to this paragraph).

The
same-to-same(-ish) points in the remote phosphor vs. white LED battle are the
color quality and efficiency. Both start with a blue (and in some cases violet)
LED emitter. Blue is most typical, and as a reference point, really good ones
can deliver a “wall plug efficiency” (WPE) of over 50%. Since our eyes
only assigns blue 1/10 to 1/3 the “brightness” that they assign to the
peak green light perception, blue photons don’t count for much in the lumen scheme
of things, but they are really handy to use to convert to longer wavelengths that
do count for a lot. The result can be 50% efficiency going in, but apparently
higher efficiency in terms of lumens (perceived brightness) coming out, as Intematix’
announcement illustrates.

The advantage falls towards white LEDs in two
key areas: Emitter size and phosphor costs. From the emitter standpoint, the rule
of optics is that the smaller the light emission source is, the smaller and closer
a narrow beam optic can be. That was a major factor in Soraa’s ability to leap
a bit ahead in the MR-16 wars, as their GaN-on-GaN approach allows them to drive
a bunch of super-tiny LEDs really-really hard, translating to lots of lumens from
a very small source. And yes, I suspect that their “tri-color phosphor”
is some special, not cheap, stuff, if it is going to stand up to the expanded
heat envelope that it is exposed to, including the greater phosphor conversion-related heat, since the “pump” is violet, which is more energetic than
blue (meaning it will have still more energy to give up in each conversion). But this
is also where the phosphor cost issue really shows up, in that the closer you
put the phosphor to the actual emitter surface, the less phosphor blend you will
need to make use of. I’ve been told that phosphor doesn’t suffer from any practical
saturation limits, so if you want to push enough blue light in there to get 1000
lumens out of a square inch, or 1000 lumens out of a square millimeter of it,
it really doesn’t care (ignoring the conversion heat issues). If the remote phosphor
and the white LED phosphor are the exact same stuff, and engineered to stand the
rigors of white LED heat, then assuming the phosphor is of some significance in
the LED’s material cost, multiplying that phosphor amount from square mm in a
direct application to the square inches of a remote approach would be a noticeable
cost factor. (I think we need to hear some stories from some phosphor folks regarding
that significance, as well as the magnitude of difference between remote versus
direct contact compositions… invitation extended phosphor folks).